db file sequential read - again

  • From: jherrick@xxxxxxx
  • To: oracle-l@xxxxxxxxxxxxx
  • Date: Tue, 13 Mar 2007 07:35:29 -0600

I know this has been discussed here before....I have a small C++ program in a
timer loop that spins around looking for the db file scattered read as well
as the db file sequential read event and logs the segments to a table on the
side. I've noticed that some of the 'sequential' read events have their P1 and
P2 parameters referring to segments marked 'TABLE' in DBA_EXTENTS. I had always
thought the 'sequential' event referred to index blocks

- Can somebody re-explain this please?

Also ... what do db file sequential reads of block #0 of a datafile
represent? This is 9.2.07 on AIX 5.3 (Asynch I/O enabled), all tablespaces
are LMT. I'm assuming these single block reads of block 0 are for
space management and therefore unavoidable?

Bonus question....this is a Siebel V7.5.3 db running in RULE mode with Indexes
up the wahzoo. I'm seeing a lot of db file sequential read waits because RBO is
using any index it finds (which we all know could be a bad thing!). I'm
thinking of putting hot indexes into their own buffer pool since I've got tons
of memory. Does this sound like a reasonable approach to you folks?

Thanks in Advance!

Jeff H


Other related posts: